In partnership with

Welcome, AI Enthusiasts! 🌟

This issue covers the latest in AI advancements: from a global AI safety treaty and Google’s new AI features like "Ask Photos" and virtual dress try-ons, to Replit’s AI Agent for app building and Salesforce's new action models. We also explore Protect AI's MLSecOps course and Midjourney's personalized Niji Mode.

Stay updated with the latest AI trends and innovations!

AI SAFETY

🌍 US, UK, EU Sign Landmark AI Safety Treaty to Protect Human Rights

Over a dozen countries, including the US, UK, and EU, have signed the Council of Europe Framework Convention on Artificial Intelligence, a legally binding treaty to ensure AI systems align with human rights, democracy, and the rule of law. Announced in Vilnius, Lithuania, the treaty outlines principles such as transparency, accountability, and privacy protection for AI lifecycle activities. Signatories must assess AI's potential risks and take measures to prevent harmful applications, marking a significant step in global AI governance.

GOOGLE

📸AI-Powered Ask Photos Feature Rolls Out in U.S.

Google has begun the U.S. rollout of its AI-powered “Ask Photos” feature, which allows users to search their Google Photos using more complex natural language queries. Powered by the Gemini AI model, this new tool goes beyond basic searches for people, places, or things, enabling users to ask questions like “best photo from each of the National Parks I visited” or “what did we order last time at this restaurant?” The AI analyzes various signals, such as lighting and photo quality, to deliver more precise results, making photo searches smarter and more intuitive.

Initially available in early access to select U.S. users, Ask Photos will eventually expand to a broader audience. The feature can even help users organize memories by identifying party themes or recalling specific events. While still in its experimental phase under Google Labs, the company emphasizes that the feature complies with its AI Principles, ensuring that private data remains secure and will not be used for ad targeting.

👗AI-Powered Virtual Try-On for Dresses

Google has expanded its AI-powered virtual try-on tool to include dresses, allowing shoppers to visualize how styles from brands like SIMKHAI and Staud will look on models that match their body type. Using the VTO-UNet Diffusion Transformer, the tool preserves body details while showcasing dress intricacies, offering a seamless and interactive shopping experience. Shoppers can even pre-order runway styles in real-time, enhancing the online shopping journey.

REPLIT

🚀 AI Agent Transforms App Building with Simple Prompts

Replit has introduced Replit Agent, an AI-powered tool that helps users create software applications using simple text descriptions, making coding more accessible to everyone. The tool allows users to describe the app they want to build in natural language, and Replit Agent takes care of the rest, coding and building web-based apps from scratch.

Key Features:

• Users can describe the app they want in plain language, and the AI generates a plan and starts coding.

• Automatically selects the right technologies and frameworks for the project.

• Builds web-based apps from scratch and allows users to interact and give feedback.

• Available on both web and mobile platforms for Replit Core and Teams subscribers.

Replit Agent removes a major barrier for new coders—setting up an integrated development environment—by handling the setup automatically. This tool empowers beginners to go from idea to a fully functional app with just a few prompts, making app development easier than ever.

SALESFORCE

💡AI Action Models Offering Lower Cost and Higher Accuracy Than Larger Rivals

Salesforce has unveiled its xLAM family of "large action models," designed to perform tasks with greater accuracy and efficiency than traditional large language models (LLMs) while significantly reducing costs. xLAM models are built for specific tool use, streamlining actions like calling apps or running Python programs instead of generating generic content. The most compact version, xLAM-1B, dubbed the "Tiny Giant," outperformed models many times its size, including OpenAI’s GPT-3.5 and Anthropic’s Claude, despite having just 1 billion parameters.

AIHEALTHTECH INSIDER: ISSUE #13

🩺AI Healthcare Revolution

Discover how Google's HeAR is revolutionizing disease detection through sound, and explore AI tools like METI for cancer diagnostics and AINU for nanoscale precision. Plus, learn about AI’s impact on predicting post-surgery opioid risks and pioneering remote robotic surgeries. Stay informed on the AI innovations transforming healthcare.

PROTECT AI

🔐Free MLSecOps Course for AI Security Certification

Protect AI has launched a free, four-part video training program called MLSecOps Foundations, designed to help organizations integrate security into their AI and machine learning workflows. The program offers 20 concise modules, covering AI security strategies and implementation using the MLSecOps framework. Led by Chief Information Security Officer Diana Kelley, the course provides participants with the knowledge to secure AI models, conduct risk assessments, and implement incident response plans, with certification upon completion.

AI EVENT

🌐Global AI Summit (GAIN) , 2024

• September 10-12, 2024

• Riyadh, Saudi Arabia

The Third Global AI Summit , GAIN 2024 is a major event set to take place from September 10-12, 2024, at the King Abdulaziz International Convention Centre in Riyadh, Saudi Arabia. Hosted by the Saudi Data & Artificial Intelligence Authority (SDAIA), this summit will bring together over 300 AI leaders from more than 100 countries.

The summit’s theme, “Now, Next, Never,” will explore the current state, future innovations, and ethical considerations of AI. Key topics include Generative AI, AI in urban life, and the ethical governance of AI. Notable speakers include Cristiano Amon (President & CEO, Qualcomm), Julie Sweet (CEO, Accenture), and Dr. Marc Raibert (Founder, Boston Dynamics).

MIDJOURNEY

🎨AI Personalizes Your Artistic Vision with Midjourney's Niji Mode

Midjourney has introduced a groundbreaking feature known as "model personalization" for its Niji (anime) models, allowing users to tailor the AI's output to their unique artistic preferences. This personalization, activated by adding --p to prompts, enables creators to refine the algorithm's understanding of their style, effectively customizing the visual narrative generated by Midjourney. This update not only enhances the user's creative control but also promises a more personalized and engaging experience in digital art creation.

Prompt for Personalized Anime Art with Midjourney's Niji Mode: "Create an image of a serene Japanese garden at twilight, with cherry blossoms in full bloom, using the personalized Niji mode for a touch of my preferred anime aesthetic. --p --niji 6"

Startup Success: Instant Access to AI Entrepreneur Courses!